Effects of Sensor Cover Damages on Point Clouds of Automotive Lidar

Abstract

Safe automated driving requires reliable perception sensors with low fault rates. Detecting perception sensor faults before path planning avoids fault propagation through the processing pipeline of automated vehicles. As the basis for further development of fault detection algorithms, the present work presents effects of damaged lidar sensor covers considering scratches, cracks, and holes. We used an automotive lidar, which provides point clouds, and calculated deviations between the lidar points on a target and an ideal plane representing the target to evaluate the effect of damaged covers. Results show that sensor cover damages have an effect on point cloud data.